Develop a framework to characterize helical TomoTherapy (HT) machine delivery errors and their clinical significance.Ten nasopharynx HT plans were edited to introduce errors in Jaw width (JW), couch speed (CS), gantry period (GP), gantry start position (GSP), multi leaf collimator leaf open times (MLC LOT). In case of MLC LOT only, both systematic and random delivery errors were investigated. Each error type was simulated independently for a range of magnitudes. Dose distributions for the clinical reference plans and the error simulated plans were compared to establish the magnitude for each error type which resulted in a change in clinical tolerance, defined as 5% variation in D95 of PTV70, D0.1cc of spinal cord, D0.1cc of brainstem and the smallest value of either a 10% or 3.6Gy dose variation in mean parotid dose.Dose variation from systematic delivery errors in JW ±0.5mm, CS ranges between -1% to 1.5%, GP ±1s, GSP ranges between -20 to 2.50 and MLC LOT random error up to 2% from the planned value relative to the clinical reference plan was within the set tolerance values for all the patient cohorts. GSP errors and the random MLC LOT errors with up to 10% standard deviation were found to be relatively insensitive compared to other delivery errors.This work has established a framework to characterize HT machine delivery errors. This framework could be applied to any patient dataset to determine clinically relevant HT QA tolerances.